from sklearn_benchmarks.report import Reporting
import pandas as pd
pd.set_option('display.max_colwidth', None)
pd.set_option('display.max_columns', None)
pd.set_option('display.max_rows', None)

scikit-learn vs. scikit-learn-intelex (Intel® oneAPI) benchmarks: perfect hyperparameters match

reporting = Reporting("sklearnex", config="config.yml")
reporting.run()

KNeighborsClassifier_brute_force: scikit-learn (1.0.dev0) vs. scikit-learn-intelex (2021.20210705.191215)

All estimators share the following hyperparameters: algorithm=brute.

	estimator	function	n_samples_train	n_samples	n_features	n_iter	mean_sklearn	stdev_sklearn	throughput	latency	n_jobs	n_neighbors	accuracy_score_sklearn	accuracy_score_sklearnex	mean_sklearnex	stdev_sklearnex	speedup	stdev_speedup	sklearn_profiling	sklearnex_profiling
0	KNeighborsClassifier_brute_force	fit	100000	100000	100	NaN	0.013	0.0	6.091	0.0	-1	100	NaN	NaN	0.068	0.0	0.193	0.0	See	See
3	KNeighborsClassifier_brute_force	fit	100000	100000	100	NaN	0.013	0.0	6.022	0.0	1	100	NaN	NaN	0.051	0.0	0.259	0.0	See	See
6	KNeighborsClassifier_brute_force	fit	100000	100000	100	NaN	0.013	0.0	6.016	0.0	-1	1	NaN	NaN	0.051	0.0	0.259	0.0	See	See
9	KNeighborsClassifier_brute_force	fit	100000	100000	100	NaN	0.013	0.0	5.974	0.0	-1	5	NaN	NaN	0.053	0.0	0.252	0.0	See	See
12	KNeighborsClassifier_brute_force	fit	100000	100000	100	NaN	0.013	0.0	5.953	0.0	1	5	NaN	NaN	0.051	0.0	0.262	0.0	See	See
15	KNeighborsClassifier_brute_force	fit	100000	100000	100	NaN	0.014	0.0	5.701	0.0	1	1	NaN	NaN	0.051	0.0	0.276	0.0	See	See
18	KNeighborsClassifier_brute_force	fit	100000	100000	2	NaN	0.006	0.0	0.286	0.0	-1	100	NaN	NaN	0.010	0.0	0.589	0.0	See	See
21	KNeighborsClassifier_brute_force	fit	100000	100000	2	NaN	0.006	0.0	0.274	0.0	1	100	NaN	NaN	0.010	0.0	0.613	0.0	See	See
24	KNeighborsClassifier_brute_force	fit	100000	100000	2	NaN	0.006	0.0	0.267	0.0	-1	1	NaN	NaN	0.009	0.0	0.638	0.0	See	See
27	KNeighborsClassifier_brute_force	fit	100000	100000	2	NaN	0.007	0.0	0.229	0.0	-1	5	NaN	NaN	0.010	0.0	0.709	0.0	See	See
30	KNeighborsClassifier_brute_force	fit	100000	100000	2	NaN	0.006	0.0	0.266	0.0	1	5	NaN	NaN	0.009	0.0	0.648	0.0	See	See
33	KNeighborsClassifier_brute_force	fit	100000	100000	2	NaN	0.005	0.0	0.320	0.0	1	1	NaN	NaN	0.009	0.0	0.531	0.0	See	See

	estimator	function	n_samples_train	n_samples	n_features	n_iter	mean_sklearn	stdev_sklearn	throughput	latency	n_jobs	n_neighbors	accuracy_score_sklearn	accuracy_score_sklearnex	mean_sklearnex	stdev_sklearnex	speedup	stdev_speedup	sklearn_profiling	sklearnex_profiling
1	KNeighborsClassifier_brute_force	predict	100000	1000	100	NaN	3.174	0.081	0.000	0.003	-1	100	0.840	0.733	0.399	0.015	7.957	0.367	See	See
2	KNeighborsClassifier_brute_force	predict	100000	1	100	NaN	0.029	0.002	0.000	0.029	-1	100	1.000	1.000	0.011	0.000	2.564	0.198	See	See
4	KNeighborsClassifier_brute_force	predict	100000	1000	100	NaN	2.431	0.030	0.000	0.002	1	100	0.840	0.636	0.392	0.004	6.197	0.098	See	See
5	KNeighborsClassifier_brute_force	predict	100000	1	100	NaN	0.024	0.001	0.000	0.024	1	100	1.000	1.000	0.013	0.002	1.833	0.253	See	See
7	KNeighborsClassifier_brute_force	predict	100000	1000	100	NaN	2.064	0.042	0.000	0.002	-1	1	0.632	0.840	0.445	0.007	4.641	0.119	See	See
8	KNeighborsClassifier_brute_force	predict	100000	1	100	NaN	0.028	0.002	0.000	0.028	-1	1	0.000	1.000	0.011	0.000	2.453	0.223	See	See
10	KNeighborsClassifier_brute_force	predict	100000	1000	100	NaN	3.103	0.043	0.000	0.003	-1	5	0.725	0.636	0.394	0.003	7.869	0.130	See	See
11	KNeighborsClassifier_brute_force	predict	100000	1	100	NaN	0.028	0.002	0.000	0.028	-1	5	1.000	1.000	0.012	0.000	2.455	0.204	See	See
13	KNeighborsClassifier_brute_force	predict	100000	1000	100	NaN	2.426	0.022	0.000	0.002	1	5	0.725	0.840	0.447	0.006	5.422	0.088	See	See
14	KNeighborsClassifier_brute_force	predict	100000	1	100	NaN	0.023	0.000	0.000	0.023	1	5	1.000	1.000	0.012	0.001	1.915	0.114	See	See
16	KNeighborsClassifier_brute_force	predict	100000	1000	100	NaN	1.434	0.011	0.001	0.001	1	1	0.632	0.733	0.401	0.007	3.577	0.067	See	See
17	KNeighborsClassifier_brute_force	predict	100000	1	100	NaN	0.025	0.002	0.000	0.025	1	1	0.000	1.000	0.012	0.001	2.090	0.262	See	See
19	KNeighborsClassifier_brute_force	predict	100000	1000	2	NaN	2.674	0.059	0.000	0.003	-1	100	0.890	0.854	0.087	0.002	30.754	1.072	See	See
20	KNeighborsClassifier_brute_force	predict	100000	1	2	NaN	0.011	0.002	0.000	0.011	-1	100	1.000	0.000	0.001	0.000	17.871	4.352	See	See
22	KNeighborsClassifier_brute_force	predict	100000	1000	2	NaN	2.291	0.119	0.000	0.002	1	100	0.890	0.823	0.084	0.001	27.315	1.464	See	See
23	KNeighborsClassifier_brute_force	predict	100000	1	2	NaN	0.005	0.002	0.000	0.005	1	100	1.000	0.000	0.001	0.000	7.284	3.519	See	See
25	KNeighborsClassifier_brute_force	predict	100000	1000	2	NaN	2.010	0.055	0.000	0.002	-1	1	0.844	0.876	0.139	0.004	14.492	0.611	See	See
26	KNeighborsClassifier_brute_force	predict	100000	1	2	NaN	0.017	0.017	0.000	0.017	-1	1	1.000	0.000	0.001	0.000	24.152	25.050	See	See
28	KNeighborsClassifier_brute_force	predict	100000	1000	2	NaN	2.919	0.170	0.000	0.003	-1	5	0.882	0.823	0.085	0.004	34.204	2.536	See	See
29	KNeighborsClassifier_brute_force	predict	100000	1	2	NaN	0.010	0.005	0.000	0.010	-1	5	1.000	0.000	0.001	0.000	16.714	8.998	See	See
31	KNeighborsClassifier_brute_force	predict	100000	1000	2	NaN	2.116	0.020	0.000	0.002	1	5	0.882	0.876	0.143	0.004	14.834	0.392	See	See
32	KNeighborsClassifier_brute_force	predict	100000	1	2	NaN	0.003	0.000	0.000	0.003	1	5	1.000	0.000	0.001	0.000	4.224	0.853	See	See
34	KNeighborsClassifier_brute_force	predict	100000	1000	2	NaN	1.182	0.031	0.000	0.001	1	1	0.844	0.854	0.086	0.001	13.670	0.396	See	See
35	KNeighborsClassifier_brute_force	predict	100000	1	2	NaN	0.002	0.000	0.000	0.002	1	1	1.000	0.000	0.001	0.000	3.272	0.813	See	See

KMeans_tall: scikit-learn (1.0.dev0) vs. scikit-learn-intelex (2021.20210705.191215)

All estimators share the following hyperparameters: algorithm=full, n_clusters=3, max_iter=30, n_init=1, tol=1e-16.

	estimator	function	n_samples_train	n_samples	n_features	n_iter_sklearn	mean_sklearn	stdev_sklearn	throughput	latency	init	adjusted_rand_score_sklearn	n_iter_sklearnex	adjusted_rand_score_sklearnex	mean_sklearnex	stdev_sklearnex	speedup	stdev_speedup	sklearn_profiling	sklearnex_profiling
0	KMeans_tall	fit	1000000	1000000	2	30	0.688	0.0	0.697	0.0	random	NaN	30	NaN	0.347	0.0	1.986	0.0	See	See
3	KMeans_tall	fit	1000000	1000000	2	30	0.683	0.0	0.703	0.0	k-means++	NaN	30	NaN	0.304	0.0	2.247	0.0	See	See
6	KMeans_tall	fit	1000000	1000000	100	30	7.837	0.0	3.063	0.0	random	NaN	30	NaN	4.161	0.0	1.883	0.0	See	See
9	KMeans_tall	fit	1000000	1000000	100	30	7.706	0.0	3.114	0.0	k-means++	NaN	30	NaN	3.979	0.0	1.937	0.0	See	See

	estimator	function	n_samples_train	n_samples	n_features	n_iter_sklearn	mean_sklearn	stdev_sklearn	throughput	latency	init	adjusted_rand_score_sklearn	n_iter_sklearnex	adjusted_rand_score_sklearnex	mean_sklearnex	stdev_sklearnex	speedup	stdev_speedup	sklearn_profiling	sklearnex_profiling
1	KMeans_tall	predict	1000000	1000	2	30	0.002	0.000	0.272	0.000	random	0.000	30	0.000	0.0	0.0	7.642	2.856	See	See
2	KMeans_tall	predict	1000000	1	2	30	0.003	0.003	0.000	0.003	random	1.000	30	1.000	0.0	0.0	14.104	15.770	See	See
4	KMeans_tall	predict	1000000	1000	2	30	0.002	0.000	0.288	0.000	k-means++	0.001	30	-0.000	0.0	0.0	7.554	3.316	See	See
5	KMeans_tall	predict	1000000	1	2	30	0.002	0.000	0.000	0.002	k-means++	1.000	30	1.000	0.0	0.0	9.548	4.654	See	See
7	KMeans_tall	predict	1000000	1000	100	30	0.002	0.000	12.381	0.000	random	0.001	30	0.002	0.0	0.0	6.532	2.297	See	See
8	KMeans_tall	predict	1000000	1	100	30	0.002	0.000	0.015	0.002	random	1.000	30	1.000	0.0	0.0	9.468	3.949	See	See
10	KMeans_tall	predict	1000000	1000	100	30	0.002	0.000	12.107	0.000	k-means++	0.003	30	0.002	0.0	0.0	6.762	2.350	See	See
11	KMeans_tall	predict	1000000	1	100	30	0.002	0.000	0.015	0.002	k-means++	1.000	30	1.000	0.0	0.0	6.268	5.218	See	See

KMeans_short: scikit-learn (1.0.dev0) vs. scikit-learn-intelex (2021.20210705.191215)

All estimators share the following hyperparameters: algorithm=full, n_clusters=300, max_iter=20, n_init=1, tol=1e-16.

	estimator	function	n_samples_train	n_samples	n_features	n_iter_sklearn	mean_sklearn	stdev_sklearn	throughput	latency	init	adjusted_rand_score_sklearn	n_iter_sklearnex	adjusted_rand_score_sklearnex	mean_sklearnex	stdev_sklearnex	speedup	stdev_speedup	sklearn_profiling	sklearnex_profiling
0	KMeans_short	fit	10000	10000	2	20	0.295	0.0	0.011	0.0	k-means++	NaN	20	NaN	0.050	0.0	5.922	0.0	See	See
3	KMeans_short	fit	10000	10000	2	20	0.096	0.0	0.033	0.0	random	NaN	20	NaN	0.134	0.0	0.713	0.0	See	See
6	KMeans_short	fit	10000	10000	100	20	1.142	0.0	0.140	0.0	k-means++	NaN	20	NaN	0.251	0.0	4.551	0.0	See	See
9	KMeans_short	fit	10000	10000	100	20	0.336	0.0	0.476	0.0	random	NaN	20	NaN	0.626	0.0	0.538	0.0	See	See

	estimator	function	n_samples_train	n_samples	n_features	n_iter_sklearn	mean_sklearn	stdev_sklearn	throughput	latency	init	adjusted_rand_score_sklearn	n_iter_sklearnex	adjusted_rand_score_sklearnex	mean_sklearnex	stdev_sklearnex	speedup	stdev_speedup	sklearn_profiling	sklearnex_profiling
1	KMeans_short	predict	10000	1000	2	20	0.002	0.000	0.139	0.000	k-means++	-0.001	20	0.002	0.001	0.0	2.582	0.939	See	See
2	KMeans_short	predict	10000	1	2	20	0.002	0.000	0.000	0.002	k-means++	1.000	20	1.000	0.000	0.0	7.781	2.947	See	See
4	KMeans_short	predict	10000	1000	2	20	0.002	0.000	0.148	0.000	random	0.001	20	0.000	0.001	0.0	3.242	0.827	See	See
5	KMeans_short	predict	10000	1	2	20	0.002	0.001	0.000	0.002	random	1.000	20	1.000	0.000	0.0	11.285	6.347	See	See
7	KMeans_short	predict	10000	1000	100	20	0.003	0.000	4.771	0.000	k-means++	0.290	20	0.316	0.001	0.0	2.296	0.292	See	See
8	KMeans_short	predict	10000	1	100	20	0.002	0.000	0.008	0.002	k-means++	1.000	20	1.000	0.000	0.0	9.420	3.887	See	See
10	KMeans_short	predict	10000	1000	100	20	0.003	0.000	4.855	0.000	random	0.344	20	0.276	0.001	0.0	2.263	0.237	See	See
11	KMeans_short	predict	10000	1	100	20	0.002	0.000	0.008	0.002	random	1.000	20	1.000	0.000	0.0	8.898	3.773	See	See

LogisticRegression: scikit-learn (1.0.dev0) vs. scikit-learn-intelex (2021.20210705.191215)

All estimators share the following hyperparameters: penalty=l2, dual=False, tol=0.0001, C=1.0, fit_intercept=True, intercept_scaling=1, class_weight=nan, random_state=nan, solver=lbfgs, max_iter=100, multi_class=auto, verbose=0, warm_start=False, n_jobs=nan, l1_ratio=nan.

	estimator	function	n_samples_train	n_samples	n_features	n_iter	mean_sklearn	stdev_sklearn	throughput	latency	class_weight	l1_ratio	n_jobs	random_state	accuracy_score	mean_sklearnex	stdev_sklearnex	speedup	stdev_speedup	sklearn_profiling	sklearnex_profiling
0	LogisticRegression	fit	1000000	1000000	100	[20]	16.362	0.0	[-0.0721086]	0.000	NaN	NaN	NaN	NaN	NaN	2.865	0.0	5.710	0.0	See	See
3	LogisticRegression	fit	1000	1000	10000	[26]	1.279	0.0	[1.62632056]	0.001	NaN	NaN	NaN	NaN	NaN	1.135	0.0	1.127	0.0	See	See

	estimator	function	n_samples_train	n_samples	n_features	n_iter	mean_sklearn	stdev_sklearn	throughput	latency	class_weight	l1_ratio	n_jobs	random_state	accuracy_score	mean_sklearnex	stdev_sklearnex	speedup	stdev_speedup	sklearn_profiling	sklearnex_profiling
1	LogisticRegression	predict	1000000	1000	100	[20]	0.000	0.0	[46.8241519]	0.0	NaN	NaN	NaN	NaN	0.535	0.000	0.0	0.804	0.309	See	See
2	LogisticRegression	predict	1000000	1	100	[20]	0.000	0.0	[0.17838488]	0.0	NaN	NaN	NaN	NaN	1.000	0.000	0.0	0.334	0.241	See	See
4	LogisticRegression	predict	1000	100	10000	[26]	0.002	0.0	[88.55720054]	0.0	NaN	NaN	NaN	NaN	0.290	0.004	0.0	0.524	0.089	See	See
5	LogisticRegression	predict	1000	1	10000	[26]	0.000	0.0	[15.93676781]	0.0	NaN	NaN	NaN	NaN	1.000	0.001	0.0	0.153	0.083	See	See

Ridge: scikit-learn (1.0.dev0) vs. scikit-learn-intelex (2021.20210705.191215)

All estimators share the following hyperparameters: alpha=1.0, fit_intercept=True, normalize=deprecated, copy_X=True, max_iter=nan, tol=0.001, solver=auto, random_state=nan.

	estimator	function	n_samples_train	n_samples	n_features	n_iter	mean_sklearn	stdev_sklearn	throughput	latency	max_iter	random_state	r2_score	mean_sklearnex	stdev_sklearnex	speedup	stdev_speedup	sklearn_profiling	sklearnex_profiling
0	Ridge	fit	1000	1000	10000	NaN	0.291	0.0	0.275	0.0	NaN	NaN	NaN	0.298	0.0	0.977	0.0	See	See
3	Ridge	fit	1000000	1000000	100	NaN	1.614	0.0	0.496	0.0	NaN	NaN	NaN	0.365	0.0	4.418	0.0	See	See

	estimator	function	n_samples_train	n_samples	n_features	n_iter	mean_sklearn	stdev_sklearn	throughput	latency	max_iter	random_state	r2_score	mean_sklearnex	stdev_sklearnex	speedup	stdev_speedup	sklearn_profiling	sklearnex_profiling
1	Ridge	predict	1000	1000	10000	NaN	0.012	0.001	6.495	0.0	NaN	NaN	0.114	0.023	0.003	0.547	0.071	See	See
2	Ridge	predict	1000	1	10000	NaN	0.000	0.000	0.707	0.0	NaN	NaN	NaN	0.000	0.000	0.604	0.375	See	See
4	Ridge	predict	1000000	1000	100	NaN	0.000	0.000	4.395	0.0	NaN	NaN	1.000	0.000	0.000	0.638	0.360	See	See
5	Ridge	predict	1000000	1	100	NaN	0.000	0.000	0.007	0.0	NaN	NaN	NaN	0.000	0.000	0.814	0.891	See	See